L'analyse de l'inflation par catégories de ménages : quelques problèmes méthodologiques

This paper revisits differences in inflation across household groups in France in the recent context of sharp price increases, primarily driven by energy and food. We identify three major sources of bias in the recent literature (Insee, CAE, OFCE, IMF, Bruegel): (i) an inadequate decomposition of contributions by consumption category, leading to an underestimation of the role of housing; (ii) the exclusion of imputed rents for owner-occupiers from price indices; and (iii) an aggregation bias in the transmission of commodity price shocks. Once these biases are corrected, the results are substantially altered. By failing to account for the consumption of housing services by owner-occupiers, the official measure of inflation tends to overestimate their inflation. This bias amounts to about 1 percentage point over two years for older households (both in absolute terms and relative to younger households), around 0.3 percentage points for households living in rural areas, and close to 1 percentage point for higher-income households. After correction, inflation heterogeneity appears much more pronounced: lower-income households experience higher inflation than higher-income households, consistent with the larger share of food and energy in their consumption baskets. This result is further reinforced when examining food inflation in greater detail, which disproportionately affects poorer households. These findings call for caution in interpreting measures of heterogeneous inflation and their implications for the design of economic policy.